• 0 Posts
  • 46 Comments
Joined 3 years ago
cake
Cake day: July 9th, 2023

help-circle
  • My experience with a Pi4B has been rock solid and plenty fast for what I’ve tried so far, at low power consumption.

    While I had problems in the beginning, it was all from the micro-SD card. People here will recommend staying away but I’ve had no problems since buying an “extreme” card. Well, starting to hit the size limit now - I can no longer kick off multiple concurrent updates because limited free space available on, I think, 16G card. Get bigger


  • If you’re starting with ha, don’t feel confined to only one.

    IMPORTANT: a local area mesh is not just a low powered way of connecting devices but is inherently local-only. Highly recommended

    The more common local area meshes include

    • Zigbee - open standard, lots of inexpensive sensors
    • Z-Wave - devices need to pay for certification but are more standard. I found more smart switches using this in my area
    • Thread - the new standard. Same frequency as Zigbee but IPv6 based. Slowly rolling out.

    The new Matter/Thread standard has support of the major players (Apple, Google, Amazon) so seems like the way to go for the future, but products are slow to roll out so you can’t count on it yet

    Personally I found the strengths of each compelling so quickly added all three of the above to my ha setup. Ha is fine with it so why limit yourself

    I follow the principle that devices must work “as expected” for my users, automation adds capabilities but does not replace them. This comes together with a focus on smart switches

    • can be used interactively just like any switch
    • continue to work as expected even if ha is down
    • typically act as routers to strengthen your local area mesh. I have switches acting as routers for Zigbee, zwave, and Thread, so all my local area meshes are solid everywhere
    • then I can automate



  • I got into each mesh technology for specific devices. Home Assistant supports them all and they seem to coexist just fine in my use case.

    I have a small to medium setup with only a few simple automations and a focus on voice control and scheduling

    Preference

    • Thread - given Apple, Amazon, and Google support and the standardization work, I expect this to be the future. Eventually. But I’m getting impatient. If I’m buying a device, I prefer Thread but usually it’s not yet
    • z-Wave - my first, and most devices. Basically this was what was most available at local stores when I started. No complaints
    • Zigbee - by far the biggest selection of simple, cheap sensors. I need to more of those
    • all too much is WiFi but I try to avoid

    But it also helps that my approach is generally switches and outlets. Hard-wired, predictable network, tend to be repeaters. I have comparatively fewer leaf nodes.

    This approach also fits in with my biggest challenge. While my house is small, it’s an older one with dense materials that blocks a lot of radio signals. For example I have no cell phone reception inside yet strong signal just out any door. My focus on switches and outlets overcome this with a repeater in every room

    So for example a few years back I got a z-wave IR blaster to control a mini-split AC because at the time I mostly used z-wave. I already had a z-wave light switch in the same room, acting as a repeater, so no worries about connectivity. Now I have both z-wave and Zigbee light switches in that room so expect both meshes to be strong for any future devices in that room




  • I’m not entirely sure why all the hate : Jenkins can do the most things the must ways. And yes, it’s so much nicer defining a pipeline with a fully functional language than an assortment of yaml files

    Actually that was my response when my company wanted to start using Gitlab ci. It only has one way of doing things so you can probably get a faster start if you had no ci, were a small company, and had simple builds. However we’re over 4,000 builds in many languages from 12 year old monoliths to modern micro services and containers…… and way too much godawful JavaScript. Do you want the quick and simple tool great for a small startup or the all powerful kitchen sink of tools?




  • Signing (intermediate) certs have been compromised before. That means a bad actor can issue fake certs that are validated up to your root ca certs

    While you can invalidate that signing cert, without useful and ubiquitous revocation lists, there’s nothing you can do to propagate that.

    A compromised signing certs, effectively means invalidating the ca cert, to limit the damage





  • I always thought this was an argument for properly racking everything. If it takes more effort, more time to remove, maybe they won’t bother.

    My understanding is that for most individuals, theft is mainly

    1. Targets of opportunity. Lock your door and make sure nothing expensive is visible
    2. Smash and Grab. The goal is to act fast and not care about what you break, so anything harder to smash (without tools) or that causes delay is good.

    I do have outside cameras but they’re not as useful as you’d think. Maybe they have some deterrent value but they’re not going to alert anyone fast enough unless they’re already in the house and you’re not going to identify anyone even if you catch a good shot of their face. If the do catch someone, perhaps the video is enough to say, yep






  • I’m actually planning to do an evaluation of a n ai code review tool to see what it can do. I’m actually somewhat optimistic that it could do this better than it can code

    I really want to sic it on this one junior programmer who doesn’t understand that you can’t just commit ai generated slop and expect it to work. This last code review after over 60 pieces of feedback I gave up on the rest and left it as he needs to understand when ai generated slop needs help

    Ai is usually pretty good at unit tests but it was so bad. Randomly started using a different mocking framework, it actually mocked entire classes and somehow thought that was valid to test them. Wasting tests on non-existent constructors no negative tests, tests without verifying anything. Most of all there were so many compile errors, yet he thought that was fine